Dedouit et al., Kramer et al and Vieth et al defined different staging methods over different sequences in knee MRI for bone age determination. Our study aims to examine three different methods in which the maturation stages of the knee epiphyses are evaluated by MRI in the same population, to compare the methods and results and evaluate them for some age thresholds and to discuss their applicability in forensic age estimation.

In this study, 597 knee images obtained using a 3.0T MR scanner were evaluated retrospectively. The T1-weighted and PD-weighted sequences were evaluated by two observers. Knee bone development was staged on the femur and tibia, and descriptive statistics were calculated for each stage according to age and sex. Descriptive statistics were presented according to the age and sex of the cases, and intra- and interobserver agreements were evaluated by the κ and κw statistics obtained by the Fleiss approach.

Significant changes were found in stages depending on age. The intra- and interobserver agreement levels were very good (κ > 0.80) for each method. The methods of Dedouit, Kramer and Vieth can be used as an additional criteria for forensic age estimation for 14, 15, 18 and 21 year thresholds.

Knee MRI stands out as an alternative modality for age estimation and it is necessary to prefer the method intended for the investigated age range with an individual-specific approach.

Our study stands out as the first study in which current classification methods were applied and compared in the same population for age estimation with knee MRI.